Parabox
Last updated
Last updated
A Parabox (Parallel Box Plot chart) shows a summary of the distributions of a set of fields from a single table. Each field is summarized based on its type: continuous (numerical) fields are summarized by box plots (or averages or dot plots), and categorical (name) fields are summarized as bubble plots. Individual cases may be displayed as lines connecting points on the "parallel axes". A Parabox is particularly useful for understanding how a subset compares to the entire population across a series of fields. Learn about each Parabox feature below:
Box Plots: for continuous data.
Continuous data is displayed by default as a box plot. A box plot summarizes the values in a field with the median (center line) and 2 quartiles above and below. Assuming that the data is normally distributed, data that would exceed 2 standard deviations in either direction is plotted outside the box plot as dots.
A numeric field may also be displayed as a dot strip, with a dot distinctly plotted for each case at a height corresponding to its value.
If the Value Type of a numeric field is Nominal, you can use bubbles to aggregate the values. Use dots to see the detailed distribution of values.
From: Chart Toolbar "Plot Style" ().
Bubble Plots: for categorical data.
Categorical fields are shown with a bubble for each category, sized by the number of rows with that value (equivalent to a bar chart).
Weighting: choose a weight for categorical fields.
All categorical fields may be weighted by a single numeric Measure field, specified in the chart configuration.
Parallel Axes: show individual rows as lines connecting parallel axes.
The "Parallel Axis" mode draws a line between each pair of adjacent columns (axes), connecting the various dimension values for each row in the data table. This helps show overall correlation between adjacent fields.
Coloring: categorical Bubble Plots colored, but no stacking!
Categorical bubbles are drawn in a color from the current scale if all Selected values in the category have the same color; otherwise the Selected portion is drawn in the Selected color. A Bubble Plot of the categorical coloring field will always appear colored.
Dot strips are also drawn based on the color scale. Box Plots are not drawn in color, but try switching to dot plot mode to see the effect.
Correlation Lines between the Parallel Axes are always colored. When colored items over-plot (as Correlation Lines normally do), they are drawn in order from low to high priority (blue to red in the Spectrum scale), so "hot" colors will be most prominent.
Study Selected sub-populations: for comparison.
If you have selected a subset of the population (possibly from another view), then that subset is highlighted in the Box Plots, and the selected portions of bubble plots are highlighted. This can be a useful way to study how a subset differs from the overall population across a number of fields.
The Advanced Property "Show Selection" can be turned off to inhibit display of the Selected sub-population in Bubble Plots and Box Plots; this can be useful if you want to focus on the Correlation Lines rather than the selected subsets.
Display of Correlation Lines for Unselected rows can be turned off to reduce clutter.
From: Chart Toolbar "Display Unselected Data"; Advanced Properties "Show Selection", "Display Unselected Axis".
Rearrange columns: to improve comparison.
Columns in the Parabox may be rearranged by clicking on the column header and dragging it to a new position. The Correlation Lines display in particular is sensitive to column ordering: a relationship between fields that would be obvious if columns were adjacent may be lost if they are separated.
Ordering each column: forward or reverse.
Column displays can be ordered by clicking in the header; options depend on the type of data. For Box Plots and Dot Strips (continuous data), ordering can be low to high or high to low.
For categorical data (Bubble Plots), options are similar to those for Bar Charts: order by by size (or reverse), by Selected amount (or reverse), by label (or reverse), or by original order.
From: Column headers.
Locate/Focus: quantitative readout
Touching the display shows the mouse location in the data space textually at the top of the column. For Box Plots, the value at the vertical location is shown. For Bubble Plots, the category is labeled and its size (Selected and total) is shown.
Focus occurs in a Box Plot or Dot Strip after hovering for one second, or by holding the SHIFT key, or when the "Label By Hover" property is set.
Focusing on an individual item with the mouse shows its values in the top row, along with a highlighted Correlation Line connecting all of the values across the parallel axes for that row of data. The focus line is drawn in the focus color, but the focus values at the top of the view are drawn in the data color.
From: Mouse; Advanced Property "Label By Hover".
Same scale: line up numeric fields.
Draw all Box Plots and Dot Strips using the same numeric range. This only makes sense if all the numeric fields have comparable ranges.
From: Advanced Property "Same Scale".
Labels and fonts: how text appears.
You can choose the font used for column headers and Bubble labels with the "Label Font" property. Labels on strips can be suppressed with the "Display Strip Labels" property.
The title for the chart and its font can also be customized.
From: Advanced Properties pane.
See also:
From: Chart Toolbar "Plot Style" ().
From: Chart Toolbar "Configure Data" ().
From: Chart Toolbar "Display Correlation Lines" (), "Configure Data" (); Advanced Property "Parallel Axis".